Boston, MA
Joined July 2015
- Tweets155
- Following367
- Followers354
- Likes820
Farhad Hormozdiari retweeted
M-REGLE is an AI method that simultaneously analyzes multiple health data streams, including ECGs and PPGs, to jointly learn rich representations and significantly boost the discovery of genetic links to disease. Learn all about it at goo.gle/4ehDwp4
ALT A 3-way Venn diagram of the genetic loci associated with cardiovascular disease previously discovered and reported in the GWAS catalog loci, loci discovered by U-REGLE, and loci discovered by M-REGLE.
Super excited to see our team's work on unsupervised phenotyping published (nature.com/articles/s41588-0……).
@tedyun and I wrote a blog explaining the work and results.
REGLE, a new method to understanding the genetic basis of organ function, leverages high-dimensional clinical data, requires no disease labels, and can incorporate expert-defined knowledge to reveal deeper insights into how our organs work. Learn more → goo.gle/3y5UdD7
ALT REGLE consists of three steps: 1) learning the lower-dimensional embeddings, 2) performing GWAS on each coordinate of the embeddings, and 3) creating a polygenic disease risk score while requiring only a small number of disease labels.
Farhad Hormozdiari retweeted
Introducing Med-Gemini, our new family of AI research models for medicine, building on Gemini's advanced capabilities. We've achieved state-of-the-art performance on a variety of benchmarks and unlocked novel applications.
#MedGemini #AIResearch #MedicalAI
Farhad Hormozdiari retweeted
Med-Gemini-Polygenic is the first LMM to predict health outcomes from genomic data converted to polygenic risk scores. It beats PRS linear models, even surprisingly predicts outcomes it wasn't trained for.
Super excited about our work in multimodal data & genetic architecture.
Utilizing multimodal AI to improve genetic analyses of cardiovascular traits medrxiv.org/cgi/content/shor… #medRxiv
Farhad Hormozdiari retweeted
Medicine is inherently multimodal, so our #AI systems must be too. Our Health AI team at Google Research is making progress with alternative approaches to introducing multimodal capabilities to medical #LLMs:
@GoogleAI
ai.googleblog.com/2023/08/mu…
Farhad Hormozdiari retweeted
📢New in @AJHGNews from McCaw et al.
📰An allelic-series rare-variant association test for candidate-gene discovery
👇
rb.gy/zx99a
Farhad Hormozdiari retweeted
Thank you so much for the invite - really looking forward to what should be a great conference.
Registration and abstract submission are now open for this great conference on #RespiratoryGenomics, organised by @Martin_Tobin @uniofleicester, Prof Ian Hall @UniofNottingham, and @Cambridge_Uni.
👉respgenomics2023.exordo.com
Farhad Hormozdiari retweeted
"Rare penetrant mutations confer severe risk of common diseases" medrxiv.org/content/10.1101/…
Interested in performing GWAS on high-dimensional clinical data and how to utilize deep learning to improve GWAS power. Check our new paper (medrxiv.org/content/10.1101/…).
We’re excited to describe new @GoogleHealth work that makes phenotyping more scalable. An unsupervised deep learning method, REpresentation learning for Genetic discovery on Low-dimensional Embeddings (REGLE), to perform GWAS on high-dimensional clinical data.
Farhad Hormozdiari retweeted
Deep learning model improves COPD risk prediction and gene discovery | Nature Genetics bit.ly/undefined #ai #ml #dl
Super excited that our paper is out! Amazing collaboration with @BrighamWomens, @harvardmed, @Northeastern , and @IUMedSchool.
@GoogleAI @GoogleResearch
Our recent @GoogleHealth AI Genomics work showing that deep learning approaches identify new genetic loci and improve risk models for chronic obstructive pulmonary disease (COPD) was published today in @NatureGenet !
Paper: rdcu.be/c96zx
Code: github.com/Google-Health/gen…
Super excited to present our deep learning works in health at AGBTAg as a keynote. #GoogleHealth #GoogIeResearch #deeplearning
Meet our new #AGBTAg speaker: @FHormozdiari 👋
- Research scientist in the HealthAI team at @Google Research
- He leverages deep learning and statistical genetics to understand plant genomics and improve crop resistance.
- PhD from UCLA
Learn more: agbt.org/events/agbt-ag/spea…
Farhad Hormozdiari retweeted
Introducing DeepNull by McCaw et al. (special shout out to @FHormozdiari and @acarroll_ATG)
#deeplearning #gwas #genomics
Farhad Hormozdiari retweeted
SVDSS combines multiple strategies to improve detection of structural variation using low-error long-read sequencing data. @l_denti @ParsoaKhorsand @DavisCompGen @BonizzoniPaola @PangaiaProject @RayanChikhi
nature.com/articles/s41592-0…
Farhad Hormozdiari retweeted
Our work on Structural Variations discovery from accurate long reads (@PacBio HiFi) is out (nature.com/articles/s41592-0…). Joint work with @l_denti @ParsoaKhorsand @BonizzoniPaola @RayanChikhi @DavisCompGen
Farhad Hormozdiari retweeted
Our recent work on SV calling using HiFi @PacBio reads is out!
Our work on Structural Variations discovery from accurate long reads (@PacBio HiFi) is out (nature.com/articles/s41592-0…). Joint work with @l_denti @ParsoaKhorsand @BonizzoniPaola @RayanChikhi @DavisCompGen
Interested in ML based phenotyping. Check our paper.
An Empirical Study of ML-based Phenotyping and Denoising for Improved Genomic Discovery biorxiv.org/cgi/content/shor… #bioRxiv
Farhad Hormozdiari retweeted
Excited to share new phenotyping methods for COPD, improving COPD GWAS
Collaborative work with IU and BWH, co-authors @cosentiyes, B Besaz , @babak_alipanahi, Z McCaw, last author @FHormozdiari, eng manager C McLean.
Paper: medrxiv.org/content/10.1101/…
Code: github.com/Google-Health/gen…
Farhad Hormozdiari retweeted
Double Fun In March!
March 29th at 12:00 EDT Cory Mclean & @FHormozdiari will discuss "DeepNull models non-linear covariate effects to improve phenotypic prediction and association power." (Nat Commun 2022)
#IGESJournalClub
Register Here:
harvard.zoom.us/meeting/regi…